Slower job growth raises the cost of a weak workforce plan

The US employment outlook is for growth, at a slower pace. BLS projects 5.9 million additional jobs between 2025 and 2035, a 3.5% increase compared with 10.9% in the previous decade. Josh Bersin reads that outlook as a reason for employers to improve productivity and redesign work.

It would be a mistake to turn that national projection into a headcount target for an individual business. An organization can face difficulty filling a critical role while employment grows elsewhere. It can also invest in AI and discover that the work it hoped to automate was never the constraint on output.

Before putting an AI saving into a workforce plan, follow a piece of work through to completion. Suppose a service team can prepare a response faster, but an experienced employee still has to resolve the exception behind it. Faster preparation may help the customer. It does not automatically release the experienced employee’s capacity or remove the need to develop another person who can handle that exception.

The staffing assumption should identify whose time becomes available and what the organization can do with it. Small savings scattered across many roles may be useful without creating a vacant position. The value might be shorter waits, more attention to difficult cases or time to coach a new colleague. Calling every saving a headcount reduction hides those choices.

For a priority workflow, I would ask the business owner and workforce planner to agree on the outcome before changing the staffing model. Count the review and correction work as well as the task the tool performs. Then examine whether the remaining work fits the skills and authority of the people expected to carry it.

A capability gap should lead to a defined development requirement. A delay caused by an approval queue needs a different response. Training everyone on the tool can consume the same scarce employee time without changing either problem.

These projections also leave room for hiring. An employer may need more people in one occupation while redesigning tasks in another. The national figures do not tell a Canadian organization which roles will be available locally, and they cannot establish the benefit of a particular AI deployment. Those decisions need local demand and operating evidence.

Use a bounded trial to replace the most important assumption in the plan with something observed. If accepted work increases and the human effort remains manageable, the organization has a basis for changing roles or investment. If review work expands, revise the design before booking the saving. A workforce plan becomes more useful when it shows how the promised capacity will actually become available.

Correction, October 8: an earlier version described a smaller US workforce. BLS projects higher total employment with slower growth; the headline and analysis now reflect that distinction.